Why Network Refreshes Take 3-4 Years – Part 1

by Sorell Slaymaker

Today’s networks are very complex and enterprises have been conservative in making any major changes, for fear of causing an outage that disrupts the business.  Most network projects are tactical in nature to meet short-term needs such as adding another cloud provider and putting in firewalls in the internal network to improve security.  This leads to further network complexity, higher costs, and a fragile environment that can break easily.  Every so often, the network needs to be redesigned from the ground up.

CIO’s and business leaders are busy and will only decide to refresh the network if they are experiencing a lot of pain such as outages, poor application performance, or network hacks/attacks.  Once a decision is made to refresh the network, the process starts.

This is the standard process that enterprises go through to upgrade their network:calendar

  1. Get Resources – 3-6 months – The existing network team still needs to manage the day-day needs of the business and to fix on-going outages while supporting new applications. Hiring a team can take upwards of a year since these resources are scares in the market, and bringing in contractors or managed service partners means that a lot of the knowledge will walk out the door at some point.
  2. Define requirements – 2-4 months – Interview business leaders, IT architects, network managers, and IT operations for what the network requirements are currently and in the future. While this can appear at a high level to be fair easy, every company has different requirements and different priorities, which makes this step very important, time consuming, and feeds into the next step.
  3. Issue RFPs – 6-12 months – Issue RFPs to carriers and network equipment providers to get a sense of the right services, equipment, and professional services. This process can be as much political as technical. Select one-two network service providers and one-four network equipment providers (routers, firewalls, IPS, WAN optimization).
  4. Create an Architecture – 3-6 months– Architecture creates a blueprint that all stake holders can understand for a design that has competing priorities. Cost, reliability, performance, security, and operational simplicity are competing priorities and gaining consensus takes time for people to understand the trade-offs and come to consensus.
  5. Pilot – 3-4 months – Test and pilot services and products to ensure they deliver as advertised. This is especially important for new technologies such as SD-WAN.  Besides making it work, ensuring the administration and operational troubleshooting and reporting tools are in place.  Many organizations do this step last, not first, and experience a lot of pain in their migration.
  6. Implementation – 9-18 months – This always takes longer than planned because:
    1. Change – The business requirements, leadership, and markets change on a regular basis. Enterprises are notoriously bad on multi-year projects with priorities, re-organizations, and budgets changing each year.
    2. Outages – Impacts to business from outages can grind a project to a halt and require the review of everything – architecture, vendors, staff, operations model, …
    3. Resources – Because of the complexity and hardware centric nature of networks, they require a high touch implementation with both central and on-site resources. Coordinating resources on project plans that can fluctuate is a nightmare and most upgrades require 50% more resources than were planned.
  7. Operations – 3-6 months – Getting the day to day processes, reporting, and change management with the new infrastructure and handing it off the existing IT operations team to take the support calls.
  8. Retiring the old network – 3-6 months – While this may sound obvious, fully retiring something within enterprise IT is difficult. There is always one feature or application that is better on the legacy environment due to years of tuning.  Plus, taking out the old equipment and finding things that were not identified in the discovery and requirements process.

Some of these processes can be done in parallel, but having been involved in four major network upgrades for Fortune 100 companies, and guiding many other companies through this process as a Gartner analyst, I have not seen it done in under 3 years.  Going forward, this legacy model is not sustainable.

One last interesting observation is that companies rarely make major network changes to cut operating costs unless there is a recession.  With the current booming U.S. economy, most CIOs are focused on growing top line revenue which means spending IT money to improve reliability, performance, and security.

Networks are not unique, most IT infrastructure takes 3-4 years for major upgrades.  This is why the cloud is appealing.  But this legacy model is not sustainable.  In part 2 of this blog, we will address how major network upgrades in the future can be done in less than one year.